RE: Web annotations for physical texts

Well, strictly speaking, I would argue that the Web Annotation Data Model does not support annotation of a resource that "isn’t viewable directly over the web" (from Karl's original email).  But it depends a little on what you mean by viewable over the web and why you want to use the Web Annotation Data Model for this use case.

The introduction to 3.2 says, "The Target resource is always an External Web Resource." Elsewhere the Model says, "There MUST be 1 or more target relationships associated with an Annotation" (Section 3.1), and "External Web Resources are dereferencable from their IRI" (from 1.4, Terminology).

Given these statements, one could still argue, I suppose, that dereferencable is different than viewable, but my sense is that trying to use the Web Annotation Data Model for describing and instantiating annotations of analog resources that have no digital surrogates of any sort (not even an RDF graph) is at best a bit of an edge case.

On the other hand, the Web Annotation Data Model provides (is intended to provide) a model for describing annotations of web resources in a manner that gives such annotations identity on the web and a measure of independence from the resource being annotated  (e.g., so they can be referenced and stored independently). It provides semantics for interoperability, allowing the sharing and reuse of annotations recorded in conformance to the Data Model.

And I would agree that such functionality likely could be useful in Karl's use case. But as Karl notes it requires extending the Web Annotation Selector model to work for analog resources with no digital surrogates, and as Steven notes there is a price to pay for trying to use Web Annotation in this way.  The CTS and Pelagios work cited is relevant and may suggest other approaches or at least ways to extend the Web Annotation Selector model if you do decide it worthwhile to extend Web Annotation for your use case.

Finally, it is worth noting that the Web Annotation Data Model is generally not seen on its own as a strict prescription for backend design, so using your own notation for your specific application makes sense in any case - regardless of whether you later find it useful to map your annotations to an extended version of the Web Annotation Data Model to facilitate sharing / aggregation of these annotations, etc.

Hope this adds a little more clarity (or at least reinforces your initial conclusions). I for one will be interested in hearing more about outcomes if you do make use of the Web Annotation Data Model in whatever way for your project.

-Tim Cole
University of Illinois at Urbana-Champaign


From: Karl Grossner <karl@kgeographer.org>
Sent: Tuesday, May 21, 2019 10:36 AM
To: Steven Harms <sgharms@stevengharms.com>
Cc: public-openannotation@w3.org
Subject: Re: Web annotations for physical texts

Thanks for this, Steven. Summarizing my survey of the situation...

I was a little surprised to discover this wording in the Web Annotation model spec:
“Resource: An item of interest that MAY be identified by an IRI”
“Web Resource: ...Web Resources MAY be dereferencable via their IRI.”

Given those definitions, passages and images in works and specific text instances are potentially annotatable. All that’s required is a standard form for the IRIs. And then, a selector for the Web Annotation model that identifies a page and/or an image on a page.

The most applicable thing I’ve found for IRIs is Cite Architecture CTS [1]. Documentation examples for parts of a text are all for ranges of lines. Apparently, one can define a TextInventory schema that defines other ways to refer to chapters, pages, and images, but I could find no specs or examples for that.

There is a further question of what response a system managing those IRIs should provide. All of these issues were tackled recently by a Pelagios-funded working group, “Linked Texts” [2][3]. They made progress I think, but their final report title is “Linked Texts WG: More Questions than Answers.”

For the moment, the WHGazetteer project will gather references to pages and images on pages in our own notation, then make some conversion when standards emerge for this process.

regards
Karl

[1] http://cite-architecture.org/ctsurn/overview/<https://urldefense.proofpoint.com/v2/url?u=http-3A__cite-2Darchitecture.org_ctsurn_overview_&d=DwMFaQ&c=OCIEmEwdEq_aNlsP4fF3gFqSN-E3mlr2t9JcDdfOZag&r=1SG5aXop0rkp_O1wnbIf7XweTLUSe9Z3aj8AqEIzm8c&m=hcScLmzYfns_LyNfq-y-ErKvKhW1A03gxWODuOku6lM&s=BVDN6vxgkSruyy_oo01Ve5d2OB97KbWTP-dqGocKLuk&e=>
[2] http://commons.pelagios.org/2018/05/the-linked-texts-working-group/<https://urldefense.proofpoint.com/v2/url?u=http-3A__commons.pelagios.org_2018_05_the-2Dlinked-2Dtexts-2Dworking-2Dgroup_&d=DwMFaQ&c=OCIEmEwdEq_aNlsP4fF3gFqSN-E3mlr2t9JcDdfOZag&r=1SG5aXop0rkp_O1wnbIf7XweTLUSe9Z3aj8AqEIzm8c&m=hcScLmzYfns_LyNfq-y-ErKvKhW1A03gxWODuOku6lM&s=ziLy0j0D7XiQYBS_bALW7_OsHx644CdfAErrsTjfonE&e=>
[3] http://commons.pelagios.org/2019/03/linked-texts-wg-more-questions-than-answers/<https://urldefense.proofpoint.com/v2/url?u=http-3A__commons.pelagios.org_2019_03_linked-2Dtexts-2Dwg-2Dmore-2Dquestions-2Dthan-2Danswers_&d=DwMFaQ&c=OCIEmEwdEq_aNlsP4fF3gFqSN-E3mlr2t9JcDdfOZag&r=1SG5aXop0rkp_O1wnbIf7XweTLUSe9Z3aj8AqEIzm8c&m=hcScLmzYfns_LyNfq-y-ErKvKhW1A03gxWODuOku6lM&s=oPg8awy7l7svMPxkX3iyJjlZUk0tWTipPwUUGKXsM5M&e=>


On 5/20/19, 7:54 AM, "Steven Harms" <sgharms@stevengharms.com<mailto:sgharms@stevengharms.com>> wrote:

Karl,

As you might have inferred by lack of response, the promise of a standard way to refer to physical books' content (including maps) seems to be unfulfilled as yet.

Last Fall I tried to advance this conversation, but ultimately came to the conclusion that this practical application is undocumented and not solidified [0].

You might find my general outline of explorations [1] helpful, or the last update when I documented my grinding to a halt [2]. I've attempted another stab
at it, but feel largely like attempting to use the W3C material is not going to be profitable. Many of the ideas of the spec seem good and I plan on
implementing them, but waiting on the W3C models is likely a hinderance to my shipping a useful thing.

I hope this helps avoid frustration!

Sincerely,

Steven Harms

[0]: https://lists.w3.org/Archives/Public/semantic-web/2018Aug/0061.html<https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.w3.org_Archives_Public_semantic-2Dweb_2018Aug_0061.html&d=DwMFaQ&c=OCIEmEwdEq_aNlsP4fF3gFqSN-E3mlr2t9JcDdfOZag&r=1SG5aXop0rkp_O1wnbIf7XweTLUSe9Z3aj8AqEIzm8c&m=hcScLmzYfns_LyNfq-y-ErKvKhW1A03gxWODuOku6lM&s=REM8PqJ-F5nLopz1wBTd6_1i-5YHv0dWybDrJt9j44o&e=>
[1]: https://stevengharms.com/research/semweb/<https://urldefense.proofpoint.com/v2/url?u=https-3A__stevengharms.com_research_semweb_&d=DwMFaQ&c=OCIEmEwdEq_aNlsP4fF3gFqSN-E3mlr2t9JcDdfOZag&r=1SG5aXop0rkp_O1wnbIf7XweTLUSe9Z3aj8AqEIzm8c&m=hcScLmzYfns_LyNfq-y-ErKvKhW1A03gxWODuOku6lM&s=k-AKPBOYeW29pgjl1H1Xl9rxZCamCn1A6vw7rwLRU_8&e=>
[2]: https://stevengharms.com/research/semweb-topic/2018-11-26-toward-easier-rdf/<https://urldefense.proofpoint.com/v2/url?u=https-3A__stevengharms.com_research_semweb-2Dtopic_2018-2D11-2D26-2Dtoward-2Deasier-2Drdf_&d=DwMFaQ&c=OCIEmEwdEq_aNlsP4fF3gFqSN-E3mlr2t9JcDdfOZag&r=1SG5aXop0rkp_O1wnbIf7XweTLUSe9Z3aj8AqEIzm8c&m=hcScLmzYfns_LyNfq-y-ErKvKhW1A03gxWODuOku6lM&s=Zq7JnoxPjaG13CWauq8a7uDRHc0Anln5vAcWhm9DICc&e=>

On Wed, May 8, 2019 at 1:05 PM Karl Grossner <karl@kgeographer.org<mailto:karl@kgeographer.org>> wrote:
My challenge is to use the Web Annotation model to annotate a map figure in a print book with authority identifiers for the places it refers to. The book has ISBN and OCLC identifiers and the map figures are numbered, but it isn’t viewable directly over the web.

I take it from this discussion that a standard selector to accomplish this does not exist, so I’m left to design our own. Is that right? Comments, corrections, and suggestions greatly appreciated.

Karl

--
Karl Grossner, PhD
Technical Director, World-Historical Gazetteer
University of Pittsburgh World History Center
http://whgazetteer.<https://urldefense.proofpoint.com/v2/url?u=http-3A__whgazetteer.&d=DwMFaQ&c=OCIEmEwdEq_aNlsP4fF3gFqSN-E3mlr2t9JcDdfOZag&r=1SG5aXop0rkp_O1wnbIf7XweTLUSe9Z3aj8AqEIzm8c&m=hcScLmzYfns_LyNfq-y-ErKvKhW1A03gxWODuOku6lM&s=-psDzLCx7rbZ9-FbX60K1y-Dfh0rEaVSrqFFKZgDaBk&e=>.org




--
Steven G. Harms
PGP: E6052DAF<https://urldefense.proofpoint.com/v2/url?u=https-3A__pgp.mit.edu_pks_lookup-3Fop-3Dget-26search-3D0x337AF45BE6052DAF&d=DwMFaQ&c=OCIEmEwdEq_aNlsP4fF3gFqSN-E3mlr2t9JcDdfOZag&r=1SG5aXop0rkp_O1wnbIf7XweTLUSe9Z3aj8AqEIzm8c&m=hcScLmzYfns_LyNfq-y-ErKvKhW1A03gxWODuOku6lM&s=KXLcmmgosVIValXvvo5Ux2PogXjrBmaf0DeuK62_u3I&e=>